slub.c 116 KB
Newer Older
Christoph Lameter's avatar
Christoph Lameter committed
1
2
3
4
5
6
7
/*
 * SLUB: A slab allocator that limits cache line use instead of queuing
 * objects in per cpu and per node lists.
 *
 * The allocator synchronizes using per slab locks and only
 * uses a centralized lock to manage a pool of partial slabs.
 *
Christoph Lameter's avatar
Christoph Lameter committed
8
 * (C) 2007 SGI, Christoph Lameter
Christoph Lameter's avatar
Christoph Lameter committed
9
10
11
 */

#include <linux/mm.h>
Nick Piggin's avatar
Nick Piggin committed
12
#include <linux/swap.h> /* struct reclaim_state */
Christoph Lameter's avatar
Christoph Lameter committed
13
14
15
16
17
#include <linux/module.h>
#include <linux/bit_spinlock.h>
#include <linux/interrupt.h>
#include <linux/bitops.h>
#include <linux/slab.h>
18
#include <linux/proc_fs.h>
Christoph Lameter's avatar
Christoph Lameter committed
19
#include <linux/seq_file.h>
Vegard Nossum's avatar
Vegard Nossum committed
20
#include <linux/kmemcheck.h>
Christoph Lameter's avatar
Christoph Lameter committed
21
22
23
24
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/mempolicy.h>
#include <linux/ctype.h>
25
#include <linux/debugobjects.h>
Christoph Lameter's avatar
Christoph Lameter committed
26
#include <linux/kallsyms.h>
27
#include <linux/memory.h>
Roman Zippel's avatar
Roman Zippel committed
28
#include <linux/math64.h>
Akinobu Mita's avatar
Akinobu Mita committed
29
#include <linux/fault-inject.h>
Christoph Lameter's avatar
Christoph Lameter committed
30

31
32
#include <trace/events/kmem.h>

Christoph Lameter's avatar
Christoph Lameter committed
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
/*
 * Lock order:
 *   1. slab_lock(page)
 *   2. slab->list_lock
 *
 *   The slab_lock protects operations on the object of a particular
 *   slab and its metadata in the page struct. If the slab lock
 *   has been taken then no allocations nor frees can be performed
 *   on the objects in the slab nor can the slab be added or removed
 *   from the partial or full lists since this would mean modifying
 *   the page_struct of the slab.
 *
 *   The list_lock protects the partial and full list on each node and
 *   the partial slab counter. If taken then no new slabs may be added or
 *   removed from the lists nor make the number of partial slabs be modified.
 *   (Note that the total number of slabs is an atomic value that may be
 *   modified without taking the list lock).
 *
 *   The list_lock is a centralized lock and thus we avoid taking it as
 *   much as possible. As long as SLUB does not have to handle partial
 *   slabs, operations can continue without any centralized lock. F.e.
 *   allocating a long series of objects that fill up slabs does not require
 *   the list lock.
 *
 *   The lock order is sometimes inverted when we are trying to get a slab
 *   off a list. We take the list_lock and then look for a page on the list
 *   to use. While we do that objects in the slabs may be freed. We can
 *   only operate on the slab if we have also taken the slab_lock. So we use
 *   a slab_trylock() on the slab. If trylock was successful then no frees
 *   can occur anymore and we can use the slab for allocations etc. If the
 *   slab_trylock() does not succeed then frees are in progress in the slab and
 *   we must stay away from it for a while since we may cause a bouncing
 *   cacheline if we try to acquire the lock. So go onto the next slab.
 *   If all pages are busy then we may allocate a new slab instead of reusing
Lucas De Marchi's avatar
Lucas De Marchi committed
67
 *   a partial slab. A new slab has no one operating on it and thus there is
Christoph Lameter's avatar
Christoph Lameter committed
68
69
70
71
72
73
74
75
76
77
 *   no danger of cacheline contention.
 *
 *   Interrupts are disabled during allocation and deallocation in order to
 *   make the slab allocator safe to use in the context of an irq. In addition
 *   interrupts are disabled to ensure that the processor does not change
 *   while handling per_cpu slabs, due to kernel preemption.
 *
 * SLUB assigns one slab for allocation to each processor.
 * Allocations only occur from these slabs called cpu slabs.
 *
Christoph Lameter's avatar
Christoph Lameter committed
78
79
 * Slabs with free elements are kept on a partial list and during regular
 * operations no list for full slabs is used. If an object in a full slab is
Christoph Lameter's avatar
Christoph Lameter committed
80
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
81
82
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
83
84
85
86
87
88
89
 *
 * Slabs are freed when they become empty. Teardown and setup is
 * minimal so we rely on the page allocators per cpu caches for
 * fast frees and allocs.
 *
 * Overloading of page flags that are otherwise used for LRU management.
 *
90
91
92
93
94
95
96
97
98
99
100
101
 * PageActive 		The slab is frozen and exempt from list processing.
 * 			This means that the slab is dedicated to a purpose
 * 			such as satisfying allocations for a specific
 * 			processor. Objects may be freed in the slab while
 * 			it is frozen but slab_free will then skip the usual
 * 			list operations. It is up to the processor holding
 * 			the slab to integrate the slab into the slab lists
 * 			when the slab is no longer needed.
 *
 * 			One use of this flag is to mark slabs that are
 * 			used for allocations. Then such a slab becomes a cpu
 * 			slab. The cpu slab may be equipped with an additional
102
 * 			freelist that allows lockless access to
103
104
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
105
106
107
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
108
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
109
110
 */

111
112
113
114
115
#define SLAB_DEBUG_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
		SLAB_TRACE | SLAB_DEBUG_FREE)

static inline int kmem_cache_debug(struct kmem_cache *s)
{
116
#ifdef CONFIG_SLUB_DEBUG
117
	return unlikely(s->flags & SLAB_DEBUG_FLAGS);
118
#else
119
	return 0;
120
#endif
121
}
122

Christoph Lameter's avatar
Christoph Lameter committed
123
124
125
126
127
128
129
130
131
132
133
/*
 * Issues still to be resolved:
 *
 * - Support PAGE_ALLOC_DEBUG. Should be easy to do.
 *
 * - Variable sizing of the per node arrays
 */

/* Enable to test recovery from slab corruption on boot */
#undef SLUB_RESILIENCY_TEST

134
135
136
137
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
138
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
139

140
141
142
143
144
145
146
/*
 * Maximum number of desirable partial slabs.
 * The existence of more partial slabs makes kmem_cache_shrink
 * sort the partial list by the number of objects in the.
 */
#define MAX_PARTIAL 10

Christoph Lameter's avatar
Christoph Lameter committed
147
148
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
149

150
/*
151
152
153
 * Debugging flags that require metadata to be stored in the slab.  These get
 * disabled when slub_debug=O is used and a cache's min order increases with
 * metadata.
154
 */
155
#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
156

Christoph Lameter's avatar
Christoph Lameter committed
157
158
159
160
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
161
162
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE | \
		SLAB_FAILSLAB)
Christoph Lameter's avatar
Christoph Lameter committed
163
164

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
165
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
166

167
168
169
170
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
#define MAX_OBJS_PER_PAGE	65535 /* since page.objects is u16 */

Christoph Lameter's avatar
Christoph Lameter committed
171
/* Internal SLUB flags */
Christoph Lameter's avatar
Christoph Lameter committed
172
#define __OBJECT_POISON		0x80000000UL /* Poison object */
Christoph Lameter's avatar
Christoph Lameter committed
173
174
175
176
177
178
179
180
181

static int kmem_size = sizeof(struct kmem_cache);

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

static enum {
	DOWN,		/* No slab functionality available */
182
	PARTIAL,	/* Kmem_cache_node works */
Christoph Lameter's avatar
Christoph Lameter committed
183
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
184
185
186
187
188
	SYSFS		/* Sysfs up */
} slab_state = DOWN;

/* A list of all slab caches on the system */
static DECLARE_RWSEM(slub_lock);
Adrian Bunk's avatar
Adrian Bunk committed
189
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
190

191
192
193
194
/*
 * Tracking user of a slab.
 */
struct track {
195
	unsigned long addr;	/* Called from address */
196
197
198
199
200
201
202
	int cpu;		/* Was running on cpu */
	int pid;		/* Pid context */
	unsigned long when;	/* When did the operation occur */
};

enum track_item { TRACK_ALLOC, TRACK_FREE };

203
#ifdef CONFIG_SYSFS
Christoph Lameter's avatar
Christoph Lameter committed
204
205
206
static int sysfs_slab_add(struct kmem_cache *);
static int sysfs_slab_alias(struct kmem_cache *, const char *);
static void sysfs_slab_remove(struct kmem_cache *);
207

Christoph Lameter's avatar
Christoph Lameter committed
208
#else
209
210
211
static inline int sysfs_slab_add(struct kmem_cache *s) { return 0; }
static inline int sysfs_slab_alias(struct kmem_cache *s, const char *p)
							{ return 0; }
212
213
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
214
	kfree(s->name);
215
216
	kfree(s);
}
217

Christoph Lameter's avatar
Christoph Lameter committed
218
219
#endif

220
static inline void stat(const struct kmem_cache *s, enum stat_item si)
221
222
{
#ifdef CONFIG_SLUB_STATS
223
	__this_cpu_inc(s->cpu_slab->stat[si]);
224
225
226
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
227
228
229
230
231
232
233
234
235
236
237
238
239
240
/********************************************************************
 * 			Core slab cache functions
 *******************************************************************/

int slab_is_available(void)
{
	return slab_state >= UP;
}

static inline struct kmem_cache_node *get_node(struct kmem_cache *s, int node)
{
	return s->node[node];
}

Christoph Lameter's avatar
Christoph Lameter committed
241
/* Verify that a pointer has an address that is valid within a slab page */
242
243
244
245
246
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

247
	if (!object)
248
249
		return 1;

250
	base = page_address(page);
251
	if (object < base || object >= base + page->objects * s->size ||
252
253
254
255
256
257
258
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

259
260
261
262
263
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

264
265
266
267
268
269
270
271
272
273
274
275
static inline void *get_freepointer_safe(struct kmem_cache *s, void *object)
{
	void *p;

#ifdef CONFIG_DEBUG_PAGEALLOC
	probe_kernel_read(&p, (void **)(object + s->offset), sizeof(p));
#else
	p = get_freepointer(s, object);
#endif
	return p;
}

276
277
278
279
280
281
static inline void set_freepointer(struct kmem_cache *s, void *object, void *fp)
{
	*(void **)(object + s->offset) = fp;
}

/* Loop over all objects in a slab */
282
283
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
284
285
286
287
288
289
290
291
			__p += (__s)->size)

/* Determine object index from a given position */
static inline int slab_index(void *p, struct kmem_cache *s, void *addr)
{
	return (p - addr) / s->size;
}

292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
static inline size_t slab_ksize(const struct kmem_cache *s)
{
#ifdef CONFIG_SLUB_DEBUG
	/*
	 * Debugging requires use of the padding between object
	 * and whatever may come after it.
	 */
	if (s->flags & (SLAB_RED_ZONE | SLAB_POISON))
		return s->objsize;

#endif
	/*
	 * If we have the need to store the freelist pointer
	 * back there or track user information then we can
	 * only use the space before that information.
	 */
	if (s->flags & (SLAB_DESTROY_BY_RCU | SLAB_STORE_USER))
		return s->inuse;
	/*
	 * Else we can use all the padding etc for the allocation
	 */
	return s->size;
}

316
317
318
319
320
static inline int order_objects(int order, unsigned long size, int reserved)
{
	return ((PAGE_SIZE << order) - reserved) / size;
}

321
static inline struct kmem_cache_order_objects oo_make(int order,
322
		unsigned long size, int reserved)
323
324
{
	struct kmem_cache_order_objects x = {
325
		(order << OO_SHIFT) + order_objects(order, size, reserved)
326
327
328
329
330
331
332
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
333
	return x.x >> OO_SHIFT;
334
335
336
337
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
338
	return x.x & OO_MASK;
339
340
}

341
#ifdef CONFIG_SLUB_DEBUG
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
/*
 * Determine a map of object in use on a page.
 *
 * Slab lock or node listlock must be held to guarantee that the page does
 * not vanish from under us.
 */
static void get_map(struct kmem_cache *s, struct page *page, unsigned long *map)
{
	void *p;
	void *addr = page_address(page);

	for (p = page->freelist; p; p = get_freepointer(s, p))
		set_bit(slab_index(p, s, addr), map);
}

357
358
359
/*
 * Debug settings:
 */
360
361
362
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
363
static int slub_debug;
364
#endif
365
366

static char *slub_debug_slabs;
367
static int disable_higher_order_debug;
368

Christoph Lameter's avatar
Christoph Lameter committed
369
370
371
372
373
374
375
376
377
378
379
380
381
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
	int i, offset;
	int newline = 1;
	char ascii[17];

	ascii[16] = 0;

	for (i = 0; i < length; i++) {
		if (newline) {
382
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
383
384
			newline = 0;
		}
385
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
386
387
388
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
389
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
390
391
392
393
394
395
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
396
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
397
398
399
			ascii[i] = ' ';
			i++;
		}
400
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
	}
}

static struct track *get_track(struct kmem_cache *s, void *object,
	enum track_item alloc)
{
	struct track *p;

	if (s->offset)
		p = object + s->offset + sizeof(void *);
	else
		p = object + s->inuse;

	return p + alloc;
}

static void set_track(struct kmem_cache *s, void *object,
418
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
419
{
Akinobu Mita's avatar
Akinobu Mita committed
420
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
421
422
423
424

	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
425
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
426
427
428
429
430
431
432
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
433
434
435
	if (!(s->flags & SLAB_STORE_USER))
		return;

436
437
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
438
439
440
441
442
443
444
}

static void print_track(const char *s, struct track *t)
{
	if (!t->addr)
		return;

445
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
446
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
447
448
449
450
451
452
453
454
455
456
457
458
459
}

static void print_tracking(struct kmem_cache *s, void *object)
{
	if (!(s->flags & SLAB_STORE_USER))
		return;

	print_track("Allocated", get_track(s, object, TRACK_ALLOC));
	print_track("Freed", get_track(s, object, TRACK_FREE));
}

static void print_page_info(struct page *page)
{
460
461
	printk(KERN_ERR "INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=0x%04lx\n",
		page, page->objects, page->inuse, page->freelist, page->flags);
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477

}

static void slab_bug(struct kmem_cache *s, char *fmt, ...)
{
	va_list args;
	char buf[100];

	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
	va_end(args);
	printk(KERN_ERR "========================================"
			"=====================================\n");
	printk(KERN_ERR "BUG %s: %s\n", s->name, buf);
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
Christoph Lameter's avatar
Christoph Lameter committed
478
479
}

480
481
482
483
484
485
486
487
488
489
490
491
static void slab_fix(struct kmem_cache *s, char *fmt, ...)
{
	va_list args;
	char buf[100];

	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
	va_end(args);
	printk(KERN_ERR "FIX %s: %s\n", s->name, buf);
}

static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
Christoph Lameter's avatar
Christoph Lameter committed
492
493
{
	unsigned int off;	/* Offset of last byte */
494
	u8 *addr = page_address(page);
495
496
497
498
499
500
501
502
503
504
505

	print_tracking(s, p);

	print_page_info(page);

	printk(KERN_ERR "INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
			p, p - addr, get_freepointer(s, p));

	if (p > addr + 16)
		print_section("Bytes b4", p - 16, 16);

506
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
507
508
509
510
511
512
513
514
515
516

	if (s->flags & SLAB_RED_ZONE)
		print_section("Redzone", p + s->objsize,
			s->inuse - s->objsize);

	if (s->offset)
		off = s->offset + sizeof(void *);
	else
		off = s->inuse;

517
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
518
519
520
521
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
522
523
524
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
525
526
527
528
529
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
530
	slab_bug(s, "%s", reason);
531
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
532
533
}

534
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
535
536
537
538
{
	va_list args;
	char buf[100];

539
540
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
541
	va_end(args);
542
	slab_bug(s, "%s", buf);
543
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
544
545
546
	dump_stack();
}

547
static void init_object(struct kmem_cache *s, void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
548
549
550
551
552
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
		memset(p, POISON_FREE, s->objsize - 1);
553
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
554
555
556
	}

	if (s->flags & SLAB_RED_ZONE)
557
		memset(p + s->objsize, val, s->inuse - s->objsize);
Christoph Lameter's avatar
Christoph Lameter committed
558
559
}

560
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
561
562
563
{
	while (bytes) {
		if (*start != (u8)value)
564
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
565
566
567
		start++;
		bytes--;
	}
568
569
570
571
572
573
574
575
576
577
578
579
	return NULL;
}

static void restore_bytes(struct kmem_cache *s, char *message, u8 data,
						void *from, void *to)
{
	slab_fix(s, "Restoring 0x%p-0x%p=0x%x\n", from, to - 1, data);
	memset(from, data, to - from);
}

static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
			u8 *object, char *what,
580
			u8 *start, unsigned int value, unsigned int bytes)
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
{
	u8 *fault;
	u8 *end;

	fault = check_bytes(start, value, bytes);
	if (!fault)
		return 1;

	end = start + bytes;
	while (end > fault && end[-1] == value)
		end--;

	slab_bug(s, "%s overwritten", what);
	printk(KERN_ERR "INFO: 0x%p-0x%p. First byte 0x%x instead of 0x%x\n",
					fault, end - 1, fault[0], value);
	print_trailer(s, page, object);

	restore_bytes(s, what, value, fault, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
600
601
602
603
604
605
606
607
608
}

/*
 * Object layout:
 *
 * object address
 * 	Bytes of the object to be managed.
 * 	If the freepointer may overlay the object then the free
 * 	pointer is the first word of the object.
Christoph Lameter's avatar
Christoph Lameter committed
609
 *
Christoph Lameter's avatar
Christoph Lameter committed
610
611
612
613
614
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
 * object + s->objsize
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
615
616
617
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
618
619
620
621
 * 	We fill with 0xbb (RED_INACTIVE) for inactive objects and with
 * 	0xcc (RED_ACTIVE) for objects in use.
 *
 * object + s->inuse
Christoph Lameter's avatar
Christoph Lameter committed
622
623
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
624
625
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
626
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
627
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
628
629
630
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
631
632
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
633
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
634
 *
Christoph Lameter's avatar
Christoph Lameter committed
635
636
 * If slabcaches are merged then the objsize and inuse boundaries are mostly
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
 * may be used with merged slabcaches.
 */

static int check_pad_bytes(struct kmem_cache *s, struct page *page, u8 *p)
{
	unsigned long off = s->inuse;	/* The end of info */

	if (s->offset)
		/* Freepointer is placed after the object. */
		off += sizeof(void *);

	if (s->flags & SLAB_STORE_USER)
		/* We also have user information there */
		off += 2 * sizeof(struct track);

	if (s->size == off)
		return 1;

655
656
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
657
658
}

659
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
660
661
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
662
663
664
665
666
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
667
668
669
670

	if (!(s->flags & SLAB_POISON))
		return 1;

671
	start = page_address(page);
672
	length = (PAGE_SIZE << compound_order(page)) - s->reserved;
673
674
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
675
676
677
	if (!remainder)
		return 1;

678
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
679
680
681
682
683
684
	if (!fault)
		return 1;
	while (end > fault && end[-1] == POISON_INUSE)
		end--;

	slab_err(s, page, "Padding overwritten. 0x%p-0x%p", fault, end - 1);
685
	print_section("Padding", end - remainder, remainder);
686

Eric Dumazet's avatar
Eric Dumazet committed
687
	restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
688
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
689
690
691
}

static int check_object(struct kmem_cache *s, struct page *page,
692
					void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
693
694
695
696
697
{
	u8 *p = object;
	u8 *endobject = object + s->objsize;

	if (s->flags & SLAB_RED_ZONE) {
698
		if (!check_bytes_and_report(s, page, object, "Redzone",
699
			endobject, val, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
700
701
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
702
703
704
705
		if ((s->flags & SLAB_POISON) && s->objsize < s->inuse) {
			check_bytes_and_report(s, page, p, "Alignment padding",
				endobject, POISON_INUSE, s->inuse - s->objsize);
		}
Christoph Lameter's avatar
Christoph Lameter committed
706
707
708
	}

	if (s->flags & SLAB_POISON) {
709
		if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
710
711
712
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
713
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
714
715
716
717
718
719
720
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

721
	if (!s->offset && val == SLUB_RED_ACTIVE)
Christoph Lameter's avatar
Christoph Lameter committed
722
723
724
725
726
727
728
729
730
731
		/*
		 * Object and freepointer overlap. Cannot check
		 * freepointer while object is allocated.
		 */
		return 1;

	/* Check free pointer validity */
	if (!check_valid_pointer(s, page, get_freepointer(s, p))) {
		object_err(s, page, p, "Freepointer corrupt");
		/*
Nick Andrew's avatar
Nick Andrew committed
732
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
733
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
734
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
735
		 */
736
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
737
738
739
740
741
742
743
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
744
745
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
746
747
748
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
749
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
750
751
		return 0;
	}
752

753
	maxobj = order_objects(compound_order(page), s->size, s->reserved);
754
755
756
757
758
759
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
760
		slab_err(s, page, "inuse %u > max %u",
761
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
762
763
764
765
766
767
768
769
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
770
771
 * Determine if a certain object on a page is on the freelist. Must hold the
 * slab lock to guarantee that the chains are in a consistent state.
Christoph Lameter's avatar
Christoph Lameter committed
772
773
774
775
776
777
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
778
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
779

780
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
781
782
783
784
785
786
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
787
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
788
789
				break;
			} else {
790
				slab_err(s, page, "Freepointer corrupt");
791
				page->freelist = NULL;
792
				page->inuse = page->objects;
793
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
794
795
796
797
798
799
800
801
802
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

803
	max_objects = order_objects(compound_order(page), s->size, s->reserved);
804
805
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
806
807
808
809
810
811
812

	if (page->objects != max_objects) {
		slab_err(s, page, "Wrong number of objects. Found %d but "
			"should be %d", page->objects, max_objects);
		page->objects = max_objects;
		slab_fix(s, "Number of objects adjusted.");
	}
813
	if (page->inuse != page->objects - nr) {
814
		slab_err(s, page, "Wrong object count. Counter is %d but "
815
816
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
817
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
818
819
820
821
	}
	return search == NULL;
}

822
823
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
{
	if (s->flags & SLAB_TRACE) {
		printk(KERN_INFO "TRACE %s %s 0x%p inuse=%d fp=0x%p\n",
			s->name,
			alloc ? "alloc" : "free",
			object, page->inuse,
			page->freelist);

		if (!alloc)
			print_section("Object", (void *)object, s->objsize);

		dump_stack();
	}
}

839
840
841
842
843
844
/*
 * Hooks for other subsystems that check memory allocations. In a typical
 * production configuration these hooks all should produce no code at all.
 */
static inline int slab_pre_alloc_hook(struct kmem_cache *s, gfp_t flags)
{
845
	flags &= gfp_allowed_mask;
846
847
848
849
850
851
852
853
	lockdep_trace_alloc(flags);
	might_sleep_if(flags & __GFP_WAIT);

	return should_failslab(s->objsize, flags, s->flags);
}

static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags, void *object)
{
854
	flags &= gfp_allowed_mask;
855
	kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
856
857
858
859
860
861
862
	kmemleak_alloc_recursive(object, s->objsize, 1, s->flags, flags);
}

static inline void slab_free_hook(struct kmem_cache *s, void *x)
{
	kmemleak_free_recursive(x, s->flags);

863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
	/*
	 * Trouble is that we may no longer disable interupts in the fast path
	 * So in order to make the debug calls that expect irqs to be
	 * disabled we need to disable interrupts temporarily.
	 */
#if defined(CONFIG_KMEMCHECK) || defined(CONFIG_LOCKDEP)
	{
		unsigned long flags;

		local_irq_save(flags);
		kmemcheck_slab_free(s, x, s->objsize);
		debug_check_no_locks_freed(x, s->objsize);
		local_irq_restore(flags);
	}
#endif
878
879
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
		debug_check_no_obj_freed(x, s->objsize);
880
881
}

882
/*
Christoph Lameter's avatar
Christoph Lameter committed
883
 * Tracking of fully allocated slabs for debugging purposes.
884
 */
Christoph Lameter's avatar
Christoph Lameter committed
885
static void add_full(struct kmem_cache_node *n, struct page *page)
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
{
	spin_lock(&n->list_lock);
	list_add(&page->lru, &n->full);
	spin_unlock(&n->list_lock);
}

static void remove_full(struct kmem_cache *s, struct page *page)
{
	struct kmem_cache_node *n;

	if (!(s->flags & SLAB_STORE_USER))
		return;

	n = get_node(s, page_to_nid(page));

	spin_lock(&n->list_lock);
	list_del(&page->lru);
	spin_unlock(&n->list_lock);
}

906
907
908
909
910
911
912
913
/* Tracking of the number of slabs for debugging purposes */
static inline unsigned long slabs_node(struct kmem_cache *s, int node)
{
	struct kmem_cache_node *n = get_node(s, node);

	return atomic_long_read(&n->nr_slabs);
}

914
915
916
917
918
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

919
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
920
921
922
923
924
925
926
927
928
{
	struct kmem_cache_node *n = get_node(s, node);

	/*
	 * May be called early in order to allocate a slab for the
	 * kmem_cache_node structure. Solve the chicken-egg
	 * dilemma by deferring the increment of the count during
	 * bootstrap (see early_kmem_cache_node_alloc).
	 */
929
	if (n) {
930
		atomic_long_inc(&n->nr_slabs);
931
932
		atomic_long_add(objects, &n->total_objects);
	}
933
}
934
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
935
936
937
938
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
939
	atomic_long_sub(objects, &n->total_objects);
940
941
942
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
943
944
945
946
947
948
static void setup_object_debug(struct kmem_cache *s, struct page *page,
								void *object)
{
	if (!(s->flags & (SLAB_STORE_USER|SLAB_RED_ZONE|__OBJECT_POISON)))
		return;

949
	init_object(s, object, SLUB_RED_INACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
950
951
952
	init_tracking(s, object);
}

953
static noinline int alloc_debug_processing(struct kmem_cache *s, struct page *page,
954
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
955
956
957
958
{
	if (!check_slab(s, page))
		goto bad;

959
	if (!on_freelist(s, page, object)) {
960
		object_err(s, page, object, "Object already allocated");
961
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
962
963
964
965
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
966
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
967
968
	}

969
	if (!check_object(s, page, object, SLUB_RED_INACTIVE))
Christoph Lameter's avatar
Christoph Lameter committed
970
971
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
972
973
974
975
	/* Success perform special debug activities for allocs */
	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_ALLOC, addr);
	trace(s, page, object, 1);
976
	init_object(s, object, SLUB_RED_ACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
977
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
978

Christoph Lameter's avatar
Christoph Lameter committed
979
980
981
982
983
bad:
	if (PageSlab(page)) {
		/*
		 * If this is a slab page then lets do the best we can
		 * to avoid issues in the future. Marking all objects
Christoph Lameter's avatar
Christoph Lameter committed
984
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
985
		 */
986
		slab_fix(s, "Marking all objects used");
987
		page->inuse = page->objects;
988
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
989
990
991
992
	}
	return 0;
}

993
994
static noinline int free_debug_processing(struct kmem_cache *s,
		 struct page *page, void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
995
996
997
998
999
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
1000
		slab_err(s, page, "Invalid object pointer 0x%p", object);